- Email: [email protected]
Associations of Patient Demographic Characteristics and Regional Physician Density With Early Physician Follow-Up Among Medicare Beneficiaries Hospitalized With Heart Failure
Associations of Patient Demographic Characteristics and Regional Physician Density With Early Physician Follow-Up Among Medicare Beneficiaries Hospitalized With Heart Failure Robb D. Kociol, MDa,b,*, Melissa A. Greiner, MSa, Gregg C. Fonarow, MDc, Bradley G. Hammill, MSa, Paul A. Heidenreich, MDd, Clyde W. Yancy, MDe, Eric D. Peterson, MD, MPHa,b, Lesley H. Curtis, PhDa,b, and Adrian F. Hernandez, MD, MHSa,b Early physician follow-up after a heart failure (HF) hospitalization is associated with lower risk of readmission. However, factors associated with early physician follow-up are not well understood. We identified 30,136 patients with HF >65 years at 225 hospitals participating in the Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients With Heart Failure (OPTIMIZE) registry or the Get With The Guidelines–Heart Failure (GWTG-HF) registry from January 1, 2003 through December 31, 2006. We linked these clinical data to Medicare claims data for longitudinal follow-up. Using logistic regression models with site-level random effects, we identified predictors of physician follow-up within 7 days of hospital discharge. Overall 11,420 patients (37.9%) had early physician follow-up. Patients residing in hospital referral regions with higher physician concentration were significantly more likely to have early follow-up (odds ratio 1.29, 95% confidence interval 1.12 to 1.48, for highest vs lowest quartile). Patients in rural areas (0.84, 0.78 to 0.91) and patients with lower socioeconomic status (0.79, 0.74 to 0.85) were less likely to have early follow-up. Women (0.87, 0.83 to 0.91) and black patients (0.84, 0.77 to 0.92) were less likely to receive early follow-up. Patients with greater co-morbidity were less likely to receive early follow-up. In conclusion, physician follow-up within 7 days after discharge from a HF hospitalization varied according to regional physician density, rural location, socioeconomic status, gender, race, and co-morbid conditions. Strategies are needed to ensure access among vulnerable populations to this supply-sensitive resource. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;108:985–991) Heart failure (HF) is the most frequent discharge diagnosis in Medicare beneficiaries and accounts for total annual costs to the United States health care system of ⬎$39.2 billion.1 Patients admitted to the hospital for HF are at high risk for postdischarge readmission or death. One-fourth of Medicare beneficiaries hospitalized for HF are readmitted within 30 days after discharge.2 Thus policy makers and a Duke Clinical Research Institute and bDepartment of Medicine, Duke University School of Medicine, Durham, North Carolina; cAhmanson– UCLA, Cardiomyopathy Center, Ronald Reagan UCLA Medical Center, Los Angeles, California; dPalo Alto VA Medical Center, Palo Alto, California; eBaylor Heart and Vascular Institute, Dallas, Texas. Manuscript received March 31, 2011; revised manuscript received and accepted May 12, 2011. Get With The Guidelines-Heart Failure is a program of the American Heart Association, Dallas, Texas, and is supported in part by an unrestricted educational grant from Medtronic, Minneapolis, Minnesota. OPTIMIZE-HF and GWTG-HF were previously supported by GlaxoSmithKline, Brentford, Middlesex, United Kingdom. This work was supported by Award 087512N from the American Heart Association-Pharmaceutical Roundtable and David and Stevie Spina, Dallas, Texas. Dr. Hernandez was supported by Award 0675060N from the American Heart AssociationPharmaceutical Roundtable and David and Stevie Spina. Dr. Peterson and Dr. Curtis were supported by Grant U18HS016964 from the Agency for Healthcare Research and Quality, Rockville, Maryland. *Corresponding author: Tel: 919-668-5959; fax: 919-668-7063. E-mail address: [email protected] (R.D. Kociol).
0002-9149/11/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2011.05.032
payer organizations are increasingly interested in improving care for patients with HF, decreasing readmissions, and lowering costs for Medicare beneficiaries. For example, the Centers for Medicare and Medicaid Services publicly reports hospital-level 30-day risk-standardized rates of readmission and mortality for HF.3,4 Early physician follow-up after discharge is a central component of efforts to decrease risk of readmission.5–7 Physician follow-up within 7 days of discharge decreases 30-day readmission rates in patients with HF.7 Despite mounting evidence supporting early followup, many patients do not receive it. Outpatient office visits are a supply-sensitive resource with often changing times to availability. Whether physician density or other associated factors affect access to early physician follow-up in recently discharged patients with HF is not well understood.8 Therefore, we examined data from the Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients With Heart Failure (OPTIMIZE-HF) registry and the Get With The Guidelines–Heart Failure (GWTG-HF) registry linked to inpatient and outpatient Medicare claims to describe patient-level characteristics and market factors associated with early physician follow-up after discharge from an HF hospitalization. Methods We linked Medicare inpatient claims data for January 1, 2003 through December 31, 2006 with data from the www.ajconline.org
986
The American Journal of Cardiology (www.ajconline.org)
Table 1 Baseline characteristics of study population Characteristic
Age (years) Age group (years) 65–69 70–74 75–79 ⱖ80 Women Race Black Other Medical history Anemia Atrial arrhythmia Chronic obstructive pulmonary disease Chronic renal insufficiency Coronary artery disease or ischemic heart disease Depression Diabetes mellitus Current or previous treatment for hyperlipidemia Current or previous treatment for hypertension Left ventricular systolic function Preserved systolic function Left ventricular systolic dysfunction Missing data Peripheral vascular disease Previous cerebrovascular accident or transient ischemic attack Smoker within the past year Findings on admission Hemoglobin (g/dl) Mean ⫾ SD Median (interquartile range) Serum creatinine (mg/dl) ⬍1.5 1.5–⬍2.0 ⱖ2.0 Serum sodium (mEq/L) Systolic blood pressure (mm Hg) Rural location State Medicaid buy-in Physicians per 100,000 residents Physicians per 100,000 residents in hospital referral region Quartile 1 (116–175) Quartile 2 (176–189) Quartile 3 (190–210) Quartile 4 (211–320) Year of index hospitalization 2003 2004 2005 2006
Early Follow-Up
p Value
Yes (n ⫽ 11,420)
No (n ⫽ 18,716)
79.0 (73.0–84.0)
79.0 (73.0–84.0)
0.03
1,460 (12.8%) 1,929 (16.9%) 2,596 (22.7%) 5,435 (47.6%) 5,858 (51.3%)
2,659 (14.2%) 3,364 (18.0%) 3,996 (21.4%) 8,697 (46.5%) 10,211 (54.6%)
⬍0.001 0.02 0.005 0.06 ⬍0.001
1,014 (8.9%) 10,406 (91.1%)
2,200 (11.8%) 16,516 (88.2%)
⬍0.001 ⬍0.001
1,928 (16.9%) 4,342 (38.0%) 3,014 (26.4%) 1,991 (17.4%) 6,171 (54.0%) 989 (8.7%) 4,464 (39.1%) 4,359 (38.2%) 8,148 (71.3%)
3,096 (16.5%) 6,350 (33.9%) 5,256 (28.1%) 3,464 (18.5%) 9,937 (53.1%) 1,691 (9.0%) 7,428 (39.7%) 6,754 (36.1%) 13,537 (72.3%)
0.44 ⬍0.001 0.001 0.02 0.11 0.27 0.30 ⬍0.001 0.07
5,837 (51.1%) 4,201 (36.8%) 1,382 (12.1%) 1,519 (13.3%) 1,794 (15.7%) 1,007 (8.8%)
9,146 (48.9%) 7,291 (39.0%) 2,279 (12.2%) 2,656 (14.2%) 3,003 (16.0%) 1,908 (10.2%)
⬍0.001 ⬍0.001 0.85 0.03 0.44 ⬍0.001 0.15
12.2 ⫾ 8.7 12.0 (10.7–13.4) 1.3 (1.0–1.8) 6,786 (59.4%) 2,386 (20.9%) 2,173 (19.0%) 138.0 (135.0–141.0) 140.0 (121.0–160.0) 1,940 (17.0%) 1,683 (14.7%) 191.5 (183.1–210.4)
12.1 ⫾ 3.7 12.1 (10.8–13.4) 1.3 (1.0–1.8) 10,833 (57.9%) 3,820 (20.4%) 3,930 (21.0%) 138.0 (136.0–141.0) 140.0 (120.0–160.0) 3,688 (19.7%) 3,510 (18.8%) 191.4 (177.8–210.0)
1,557 (13.6%) 3,445 (30.2%) 3,583 (31.4%) 2,835 (24.8%)
3,031 (16.2%) 5,761 (30.8%) 5,542 (29.6%) 4,382 (23.4%)
⬍0.001 0.008 0.32 ⬍0.001 ⬍0.001 0.98 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 0.26 0.001 0.005
2,676 (23.4%) 4,269 (37.4%) 1,741 (15.2%) 2,734 (23.9%)
4,608 (24.6%) 6,892 (36.8%) 2,909 (15.5%) 4,307 (23.0%)
0.02 0.33 0.49 0.07
Values are presented as median (interquartile range), number of patients (percentage), or mean ⫾ SD.
OPTIMIZE-HF and GWTG-HF registries. GWTG-HF is a continuation of the OPTIMIZE-HF under the sponsorship of the American Heart Association, and the registries had the same design, inclusion criteria, and data collection methods.9,10 Patients were eligible for inclusion in the registries if they were admitted to a hospital for an episode of worsening HF or developed significant HF symptoms during a
hospitalization for which HF was the primary discharge diagnosis. Participating institutions submitted data on consecutive eligible patients in compliance with Joint Commission and Centers for Medicare and Medicaid Services standards. The registries included hospitals from all regions of the United States ranging in type from community hospitals to academic tertiary care referral centers. The validity and
Heart Failure/Early Physician Follow-up for Heart Failure
987
Table 2 Inpatient procedures, discharge processes, and discharge medications Variable
Percutaneous coronary intervention performed during hospitalization Coronary artery bypass grafting performed during hospitalization Automated implantable cardioverter–defibrillator placement in hospitalization Right cardiac catheterization performed during hospitalization Length of stay (days), median (interquartile range) Referral to outpatient heart failure management program Discharge instructions completed  Blocker Angiotensin-converting enzyme inhibitor or angiotensin receptor blocker Antiplatelet agent Lipid-lowering agent Digoxin Diuretic Aldosterone antagonist
generalizability of the OPTIMIZE-HF registry have been described previously.11 We obtained research-identifiable Medicare inpatient, carrier, and denominator files for each patient in the study for 2003 through 2007. Denominator files contain death date and Medicare eligibility dates. We used the inpatient files to identify readmissions and the carrier files to examine early physician follow-up after discharge from the index HF hospitalization. Carrier files contain claims from noninstitutional medical providers for services covered by Medicare Part B and include Healthcare Common Procedure Coding System codes, physician specialty codes, and service dates. We used data from 2007 for hospital discharges that occurred in December 2006. This study was approved by the institutional review board of the Duke University Health System. All sites participating in OPTIMIZE-HF and GWTG-HF received local institutional review board approval, if applicable, and complied with all local regulatory and privacy guidelines. Outcome Sciences, Inc. (Cambridge, Massachusetts) served as the data collection and coordination center for GWTGHF. The Duke Clinical Research Institute (Durham, North Carolina) served as the data analysis center and entered an agreement to analyze the aggregate deidentified data for research purposes (http://www.Clinicaltrials.gov, identifier NCT00344513). We used a method that links registry records for patients ⱖ65 years with inpatient Medicare claims files based on indirect identifiers.10 Using this method we were able to link 62,311 (78%) of the 79,837 eligible hospitalizations in the OPTIMIZE-HF and GWTG-HF registries with Medicare inpatient claims. Eligible patients were enrolled in fee-forservice Medicare and were discharged alive from a fully participating site in the OPTIMIZE-HF or GWTG-HF program. For patients with multiple admissions we selected the first as the index admission. We excluded patients who were discharged to a skilled nursing facility (n ⫽ 9,166) or to hospice care (n ⫽ 804). We also excluded 1,390 patients from 143 hospitals that had ⬍25 eligible patients remaining in the cohort.
Early Follow-Up Yes (n ⫽ 11,420)
No (n ⫽ 18,716)
161 (1.4%) 56 (0.5%) 311 (2.7%) 307 (2.7%) 4 (2–6) 1,453 (12.7%) 6,959 (60.9%) 7,859 (68.8%) 7,316 (64.1%) 6,063 (53.1%) 5,005 (43.8%) 2,988 (26.2%) 9,309 (81.5%) 1,534 (13.4%)
243 (1.3%) 123 (0.7%) 623 (3.3%) 534 (2.9%) 4 (3–6) 2,166 (11.6%) 11,179 (59.7%) 12,612 (67.4%) 11,750 (62.8%) 10,468 (55.9%) 7,861 (42.0%) 4,807 (25.7%) 14,784 (79.0%) 2,336 (12.5%)
p Value
0.41 0.07 0.003 0.40 ⬍0.001 0.003 0.04 0.01 0.03 ⬍0.001 0.002 0.36 ⬍0.001 0.02
We defined early physician follow-up as an outpatient evaluation and management visit with a physician (Healthcare Common Procedure Coding System codes 992.xx through 994.xx) within 7 days after discharge from the index hospitalization. Although early follow-up can be defined from any number of time points, we chose 7 days to be consistent with the American College of Cardiology’s H2H national quality-improvement initiative. The H2H initiative involves ⬎800 centers in an effort to decrease hospital readmissions in patients with HF and myocardial infarction by 20% by 2012. Postdischarge physician follow-up within 7 days is 1 of 3 core goals of the initiative.6 The importance of physician follow-up within 7 days is also supported by evidence from a recent study in which hospitals that arranged 7-day follow-up for a larger proportion of patients with HF had lower readmission rates than other hospitals.7 We excluded emergency department visits from calculations of early follow-up rates because they were unplanned visits not reflecting a system of care. We obtained patient demographic characteristics, medical history, results of admission laboratory tests and examinations, discharge pharmacy records, and procedural information for the index hospitalization from the registry data. Patients were assigned to race categories using options available on the case-report form. We used the reported category “black” and combined all others as “other.” Variables in the analysis had low rates of “missingness” (i.e., ⬍5% of records) with the exception of left ventricular function (12.1%). For continuous variables and the variable for left ventricular function, we created categorical variables that included a category for missing values. For other dichotomous variables (i.e., smoker within previous year, discharge processes and performance measurements, and index hospitalization procedures) we imputed missing values to “no.” We used the enrollment code in the month of the index hospitalization to ascertain Medicaid eligibility, a marker of socioeconomic status.12,13 A dichotomous indicator variable for rural status was derived from rural-urban commuting area scores based on zip code of residence14 using the University of Washington classification C algo-
988
The American Journal of Cardiology (www.ajconline.org)
Table 3 Univariate and multivariable predictors of seven-day physician follow-up Variable Age group (years) 65–69 70–74 75–79 ⱖ80 Female gender Black race Medical history Atrial arrhythmia Chronic obstructive pulmonary disease Coronary artery disease or ischemic heart disease Depression Diabetes mellitus Hyperlipidemia Peripheral vascular disease Previous cerebrovascular accident or transient ischemic attack Smoker within previous year Systolic function Preserved systolic function Left ventricular systolic dysfunction Findings on admission Serum creatinine (mg/dl) ⬍1.5 1.5–⬍2.0 ⱖ2.0 Systolic blood pressure (mm Hg) ⬍120 120–⬍140 140–⬍160 ⱖ160 Missing Serum sodium (mEq/L) ⬍135 135–⬍145 ⱖ145 Hemoglobin (g/dl) ⬍9 9–⬍12 ⱖ12 Findings at discharge Referral to outpatient heart failure management program Discharge instructions completed Length of index hospitalization ⬎7 days Rural location State Medicaid buy-in Quartile of physicians per 100,000 residents in hospital referral region 1 (116–175) 2 (176–189) 3 (190–210) 4 (211–320) Year of index hospitalization 2003 2004 2005 2006 * Multivariable model includes all listed variables. CI ⫽ confidence interval; OR ⫽ odds ratio.
Unadjusted OR (95% CI)
p Value
Adjusted OR* (95% CI)
p Value
1.00 (reference) 1.04 (0.96–1.13) 1.18 (1.08–1.28) 1.11 (1.03–1.20) 0.88 (0.84–0.92) 0.76 (0.70–0.83)
0.35 ⬍0.001 0.005 ⬍0.001 ⬍0.001
1.00 (reference) 1.01 (0.93–1.11) 1.12 (1.03–1.22) 1.04 (0.96–1.13) 0.87 (0.83–0.91) 0.84 (0.77–0.92)
0.74 0.007 0.29 ⬍0.001 ⬍0.001
1.20 (1.14–1.26) 0.94 (0.89–0.99) 1.05 (1.00–1.10) 0.98 (0.90–1.07) 0.99 (0.94–1.04) 1.10 (1.05–1.16) 0.94 (0.88–1.01) 0.98 (0.92–1.05)
⬍0.001 0.02 0.07 0.68 0.62 ⬍0.001 0.12 0.53
1.17 (1.11–1.23) 0.94 (0.89–1.00) 1.03 (0.98–1.09) 0.99 (0.91–1.08) 1.01 (0.96–1.06) 1.09 (1.04–1.15) 0.94 (0.87–1.00) 0.99 (0.92–1.05)
⬍0.001 0.04 0.21 0.87 0.68 0.001 0.06 0.68
0.87 (0.80–0.94)
0.001
0.92 (0.84–1.01)
0.07
1.00 (reference) 0.91 (0.87–0.96)
0.001
1.00 (reference) 0.89 (0.85–0.94)
⬍0.001
1.00 (reference) 1.00 (0.94–1.06) 0.89 (0.84–0.95)
0.92 ⬍0.001
1.00 (reference) 0.97 (0.91–1.03) 0.88 (0.82–0.94)
0.35 ⬍0.001
1.00 (reference) 1.02 (0.95–1.09) 1.01 (0.95–1.09) 1.02 (0.95–1.09) 1.11 (0.75–1.65)
0.62 0.72 0.59 0.59
1.00 (reference) 1.02 (0.95–1.09) 1.02 (0.95–1.10) 1.05 (0.98–1.13) 1.00 (0.67–1.52)
0.66 0.54 0.19 0.98
1.00 (reference) 0.90 (0.85–0.96) 0.88 (0.76–1.01)
0.001 0.06
1.00 (reference) 0.90 (0.85–0.96) 0.89 (0.77–1.02)
0.001 0.09
1.00 (reference) 1.07 (0.96–1.20) 1.02 (0.91–1.13)
0.23 0.77
1.00 (reference) 1.05 (0.94–1.17) 0.97 (0.87–1.08)
0.40 0.59
1.16 (1.07–1.25)
⬍0.001
1.14 (1.05–1.24)
0.002
1.07 (1.01–1.13) 0.96 (0.90–1.02) 0.84 (0.78–0.90) 0.74 (0.69–0.79)
0.01 0.19 ⬍0.001 ⬍0.001
1.04 (0.98–1.10) 0.97 (0.91–1.04) 0.84 (0.78–0.91) 0.79 (0.74–0.85)
0.20 0.38 ⬍0.001 ⬍0.001
1.00 (reference) 1.23 (1.08–1.41) 1.29 (1.14–1.47) 1.33 (1.16–1.52)
0.002 ⬍0.001 ⬍0.001
1.00 (reference) 1.21 (1.06–1.38) 1.29 (1.14–1.46) 1.29 (1.12–1.48)
0.005 ⬍0.001 ⬍0.001
1.00 (reference) 1.04 (0.97–1.12) 1.00 (0.91–1.10) 1.12 (1.02–1.23)
0.22 0.99 0.02
1.00 (reference) 1.03 (0.96–1.10) 0.99 (0.89–1.09) 1.09 (0.99–1.20)
0.46 0.77 0.08
Heart Failure/Early Physician Follow-up for Heart Failure
rithm.15 We assigned each patient to a hospital referral region according to the zip code of residence.16 We assigned hospital referral regions to quartiles based on total number of physicians per 100,000 residents and linked rankings to patients by hospital referral region. For baseline patient characteristics and index hospitalization measurements we present categorical variables as frequencies with percentages and continuous variables as medians with interquartile ranges. To test for differences in early physician follow-up, we used chi-square tests for categorical variables and Kruskal–Wallis tests for continuous variables. We used univariate logistic regression models and multivariable logistic regression models with site-level random effects to examine predictors of early physician follow-up. In multivariable analysis we modeled early follow-up as a function of age, gender, race, medical history, results of admission laboratory tests and examinations, completion of discharge instructions, referral to an HF disease management program, length of stay ⬎7 days for the index hospitalization, rural location, state Medicaid buy-in, quartile of physicians per 100,000 residents in the hospital referral region, and year of index hospitalization. We used a similar approach to examine unadjusted relations between covariates and early physician follow-up. In the primary analysis we included all patients in the study population. In a sensitivity analysis we excluded patients who died or were readmitted within 7 days after discharge from the index hospitalization. In a second sensitivity analysis we determined unadjusted and adjusted relations between covariates and early physician follow-up defined as 14 days including and excluding patients who died or were readmitted at 14 days. We used SAS 9.2 (SAS Institute, Cary, North Carolina) for all analyses. Results The study population included 30,136 patients from 225 hospitals. Within 7 days of discharge from the index HF hospitalization, 11,420 patients (37.9%) had a follow-up visit with a physician. Table 1 lists baseline demographic characteristics of the study population stratified by early follow-up status. Median age in the 2 groups was 79 years. Compared to the early follow-up cohort, the cohort of patients without early follow-up had a larger proportion of black patients and women. The early follow-up cohort had higher rates of preserved systolic function and history of atrial arrhythmias. The early follow-up cohort also had lower rates of chronic obstructive pulmonary disease and chronic renal insufficiency. A larger proportion of patients with HF in the early follow-up cohort resided in a hospital referral region with a higher concentration of physicians. More patients in the cohort without early follow-up lived in a rural location and had lower socioeconomic status. Table 2 presents inpatient procedures, discharge process measurements, and discharge medications stratified by early follow-up status. In general the 2 cohorts were similar with respect to findings at discharge and procedures performed. However, larger proportions of patients in the early follow-up group were referred to an outpatient HF disease management program and had discharge instructions completed. The early follow-up cohort also had higher rates of
989
discharge with evidence-based HF medications including angiotensin-converting enzyme inhibitors or angiotensin receptor blockers,  blockers, and aldosterone antagonists. A larger proportion of the early follow-up cohort was discharged on diuretics and lipid-lowering agents. The early follow-up group had a smaller proportion of patients discharged with antiplatelet medications. Table 3 lists unadjusted and adjusted associations with early physician follow-up. After adjustment for baseline characteristics of the index hospitalization, we observed increased odds of early follow-up by quartile of number of physicians per 100,000 patients in the hospital referral region. Patients residing in hospital referral regions in the top quartile of physician density had greater odds of early follow-up compared to patients residing in regions in the lowest quartile. Moreover, patients living in rural areas and patients with lower socioeconomic status were less likely to have early physician follow-up. We found no significant improvement in crude rates or adjusted odds of early follow-up from 2003 through 2006. After multivariable adjustment for baseline characteristics of the study population, the odds of early follow-up were 13% lower in women compared to men and 16% lower in black patients compared to patients of other races. In addition, we found statistically significant independent associations between early follow-up and history of hyperlipidemia, history of atrial arrhythmias, and referral to an outpatient HF disease management program. Patients with left ventricular dysfunction, chronic obstructive pulmonary disease, or chronic renal insufficiency had lower odds of early follow-up. In a sensitivity analysis we repeated the primary analysis but excluded patients from the cohort who died or were readmitted within 7 days of discharge from the index hospitalization. This analysis did not significantly change the findings for predictors of early follow-up. In a second sensitivity analysis we changed our definition of early physician follow-up to 14 days after discharge including and excluding the population of patients who died or were readmitted at 14 days. Overall the associations and effect sizes were similar to the 7-day model. There were 2 exceptions. First, length of stay ⬎7 days was associated with decreased odds of early follow-up in the model that included patients who died or were readmitted within 14 days. Second, completed discharge instructions were associated with increased odds of early follow-up. Discussion To our knowledge, ours is the largest analysis to date of factors associated with early physician follow-up after discharge from an HF admission. The study yielded several important findings. First, patients who reside in hospital referral regions with low physician density were less likely to receive early physician follow-up. Second, overall rate of early physician follow-up was low (38%). Third, gender and race were associated with early follow-up. Fourth, patients at high risk for readmission such as those with kidney disease and chronic obstructive pulmonary disease were less likely to receive early follow-up.
990
The American Journal of Cardiology (www.ajconline.org)
The definition of early physician follow-up as 7 days after discharge is consistent with current quality-improvement initiatives and previous research. However, to determine whether the same factors were associated with early physician follow-up if we used a different cut point, we repeated the analysis using a 14-day cut point instead of 7 days and found similar associations and effect sizes. Our finding of higher rates of early follow-up in patients residing in hospital referral regions with more physicians per capita calls attention to provider access as a factor that influences the transition from inpatient to outpatient care. To our knowledge this is the first study to identify a correlation between physician supply and access to HF care. One might argue that physician density is simply a marker for something else—notably, rural location or socioeconomic status. However, we adjusted for rural location and socioeconomic status (based on Medicaid buy-in) and physician density remained significant. Moreover, patients in rural regions and from a lower socioeconomic status were less likely to receive early physician follow-up. Other studies have found geographic variations in subspecialty physician supply.17 For example, areas with lower physician density have lower rates of primary care visits and early follow-up for newborns.18 Whether physician density is directly associated with lower rates of follow-up or is a marker of other socioeconomic factors is uncertain. A recent study used a multivariable model to determine predictors of myocardial infarction and HF hospitalization rates. After adjustment for other socioeconomic variables (i.e., unemployment rate, poverty rate, median income, rural location, distance to hospital, and education level), physician density was not a significant predictor.19 However, outpatient physician visits are supply-sensitive and are thought to play a role in variations in regional health care resource use.8 Providers with larger patient panels will have a more difficult time fitting patients into their clinic after a hospital discharge in a timely fashion. In comparison, physicians with more clinic slots to fill have an incentive to schedule more patients.20 Whether increased supply-sensitive care generally for patients with HF improves overall outcome is unclear. However, our analysis suggests that increased provider access may promote early physician follow-up, which has been associated with decreased short-term readmission rates. Consistent with previous studies in other areas of medicine, black race was associated with lower odds of early physician follow-up.21–24 This finding may help to explain why previous research has shown a greater risk of readmission in black patients after hospitalization for HF.2,25,26 Similarly, women in our study were less likely to have early physician follow-up, which may contribute to higher risks for HF readmission in women.24 Therefore, improving transitional care and access to follow-up for these groups requires increased attention. We also found that patients who were at greater clinical risk for readmission or death because of left ventricular systolic dysfunction, increased serum creatinine, or chronic obstructive pulmonary disease were less likely to receive early follow-up.27–29 The explanation for this seemingly paradoxical finding of lower rates of early follow-up among subsets of higher-risk more complicated patients is unclear.
Sicker patients may have more difficulty arranging physician visits or may tend to be referred to specialists with longer wait times for visits. It is also possible that higherrisk patients with more co-morbid conditions are more likely to be readmitted or to die within 7 days of discharge and therefore do not have an opportunity to present for early follow-up. However, findings from the sensitivity analysis that excluded patients who died or were readmitted within 7 days of discharge were similar to findings from the primary analysis. Atrial arrhythmias stood out as a co-morbid condition that resulted in greater odds of early follow-up, possibly related to a need in these patients for early frequent international normalized ratio testing after discharge and physician awareness of the potentially catastrophic consequences of subtherapeutic or supratherapeutic international normalized ratio levels. We also did not find a meaningful temporal change in rates of early follow-up and year of the index hospitalization was not an independent predictor of early follow-up. This finding highlights 1 of several domains in the care of patients with acute HF who have not shown improvement over time. Our finding that enrollment in an HF disease management program after discharge is associated with early physician follow-up offers a potential intervention for the aforementioned risk/follow-up paradox. Enrollment in a disease management program may provide closer monitoring of sicker patients, thus improving adherence to follow-up appointments. Further study of who benefits most from such programs will be important. Our study has several limitations. First, analysis was restricted to patients in hospitals participating in qualityimprovement registries, which may not be generalizable to other hospitals. A previous analysis comparing registry data to the larger population found the populations to be similar.11 Second, the study included only fee-for-service Medicare beneficiaries and may not generalize to patients enrolled in Medicare managed care. Third, patients may have received follow-up care from nurse practitioners, physician assistants, home health visits, or remote monitoring that we did not capture in this analysis. Fourth, the analysis did not account for unmeasured confounders that may influence access to care. Acknowledgment: Damon M. Seils, MA (Duke University) provided editorial assistance and prepared the report. Mr. Seils did not receive compensation for his assistance apart from his employment at the institution where the study was conducted. 1. Lloyd-Jones D, Adams RJ, Brown TM, Carnethon M, Dai S, De Simone G, Ferguson TB, Ford E, Furie K, Gillespie C, Go A, Greenlund K, Haase N, Hailpern S, Ho PM, Howard V, Kissela B, Kittner S, Lackland D, Lisabeth L, Marelli A, McDermott MM, Meigs J, Mozaffarian D, Mussolino M, Nichol G, Roger VL, Rosamond W, Sacco R, Sorlie P, Roger VL, Thom T, Wasserthiel-Smoller S, Wong ND, Wylie-Rosett J; American Heart Association, Statistics Committee and Stroke Statistics. Heart disease and stroke statistics—2010 update: a report from the American Heart Association. Circulation 2010;121(suppl):e46 – e215. 2. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009; 360:1418 –1428.
Heart Failure/Early Physician Follow-up for Heart Failure 3. Medicare Payment Advisory Commission (MedPAC). Report to the Congress: promoting greater efficiency in Medicare, June 2007. Available at: http://www.medpac.gov/documents/Jun07_EntireReport.pdf. Accessed August 25, 2010. 4. US Department of Health and Human Services. Hospital compare. Available at: http://www.hospitalcompare.hhs.gov. Accessed December 6, 2009. 5. Hauptman PJ, Rich MW, Heidenreich PA, Chin J, Cummings N, Dunlap ME, Edwards ML, Gregory D, O’Connor CM, Pezzella SM, Philbin E; Heart Failure Society of America. The heart failure clinic: a consensus statement of the Heart Failure Society of America. J Card Fail 2008;14:801– 815. 6. H2H National Quality Improvement Initiative. Available at: http:// h2hquality.org. Accessed December 26, 2009. 7. Hernandez AF, Greiner MA, Fonarow GC, Hammill BG, Heidenreich PA, Yancy CW, Peterson ED, Curtis LH. Relationship between early physician follow-up and 30-day readmission among Medicare beneficiaries hospitalized for heart failure. JAMA 2010;303:1716 –1722. 8. Fisher ES, Wennberg JE. Health care quality, geographic variations, and the challenge of supply-sensitive care. Perspect Biol Med 2003; 46:69 –79. 9. Fonarow GC, Abraham WT, Albert NM, Gattis WA, Gheorghiade M, Greenberg B, O’Connor CM, Yancy CW, Young J. Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF): rationale and design. Am Heart J 2004;148:43–51. 10. Hammill BG, Hernandez AF, Peterson ED, Fonarow GC, Schulman KA, Curtis LH. Linking inpatient clinical registry data to Medicare claims data using indirect identifiers. Am Heart J 2009;157:995–1000. 11. Curtis LH, Greiner MA, Hammill BG, DiMartino LD, Shea AM, Hernandez AF, Fonarow GC. Representativeness of a national heart failure quality-of-care registry: comparison of OPTIMIZE-HF and non-OPTIMIZE-HF Medicare patients. Circ Cardiovasc Qual Outcomes 2009;2:377–384. 12. Section 3. Dual-eligible beneficiaries. In: Medicare Payment Advisory Commission (MedPAC). A Data Book: Healthcare Spending and the Medicare Program, June 2010. Available at: http://www.medpac.gov/ chapters/Jun10DataBookSec3.pdf. Accessed March 31, 2011. 13. Section 8. How do dual eligible Medicare Benes compare to non-dual eligible Medicare Benes? In: Medicare Current Beneficiary Survey (MCBS) Data Tables, 1997. Available at: http://www.cms.gov/MCBS/ Downloads/CNP_1997_summary8.pdf. Accessed March 31, 2011. 14. US Department of Agriculture, Economic Research Service. Ruralurban commuting area codes. Available at: http://www.ers.usda.gov/ Data/RuralUrbanCommutingAreaCodes/. Accessed March 31, 2011. 15. WWAMI Rural Health Research Center. RUCA data: using RUCA data. Available at: http://depts.washington.edu/uwruca/ruca-uses.php. Accessed March 31, 2011. 16. The Dartmouth Atlas of Health Care. Available at: http://www. dartmouthatlas.org/data/download.shtm. Accessed February 18, 2010.
991
17. Onega T, Duell EJ, Shi X, Wang D, Demidenko E, Goodman D. Geographic access to cancer care in the U.S. Cancer 2008;112:909 – 918. 18. Guttmann A, Shipman SA, Lam K, Goodman DC, Stukel TA. Primary care physician supply and children’s health care use, access, and outcomes: findings from Canada. Pediatrics 2010;125:1119 –1126. 19. Harris DE, Aboueissa AM, Hartley D. Myocardial infarction and heart failure hospitalization rates in Maine, USA—variability along the urban-rural continuum. Rural Remote Health 2008;8:980. 20. Wennberg JE. Unwarranted variations in healthcare delivery: implications for academic medical centres. BMJ 2002;325:961–964. 21. Sonel AF, Good CB, Mulgund J, Roe MT, Gibler WB, Smith SC, Jr., Cohen MG, Pollack CV Jr, Ohman EM, Peterson ED, CRUSADE Investigators; CRUSADE Investigators. Racial variations in treatment and outcomes of black and white patients with high-risk non-STelevation acute coronary syndromes: insights from CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines?). Circulation 2005;111:1225–1232. 22. Skinner J, Chandra A, Staiger D, Lee J, McClellan M. Mortality after acute myocardial infarction in hospitals that disproportionately treat black patients. Circulation 2005;112:2634 –2641. 23. Blomkalns AL, Chen AY, Hochman JS, Peterson ED, Trynosky K, Diercks DB, Brogan GX, Jr., Boden WE, Roe MT, Ohman EM, Gibler WB, Newby LK; CRUSADE Investigators. Gender disparities in the diagnosis and treatment of non–ST-segment elevation acute coronary syndromes: large-scale observations from the CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes with Early Implementation of the American College of Cardiology/American Heart Association Guidelines) National Quality Improvement Initiative. J Am Coll Cardiol 2005;45:832– 837. 24. Howie-Esquivel J, Dracup K. Effect of gender, ethnicity, pulmonary disease, and symptom stability on rehospitalization in patients with heart failure. Am J Cardiol 2007;100:1139 –1144. 25. Philbin EF, DiSalvo TG. Influence of race and gender on care process, resource use, and hospital-based outcomes in congestive heart failure. Am J Cardiol 1998;82:76 – 81. 26. Rathore SS, Foody JM, Wang Y, Smith GL, Herrin J, Masoudi FA, Wolfe P, Havranek EP, Ordin DL, Krumholz HM. Race, quality of care, and outcomes of elderly patients hospitalized with heart failure. JAMA 2003;289:2517–2524. 27. Patel UD, Greiner MA, Fonarow GC, Phatak H, Hernandez AF, Curtis LH. Associations between worsening renal function and 30-day outcomes among Medicare beneficiaries hospitalized with heart failure. Am Heart J 2010;160(suppl):132.e1–138.e1. 28. Krumholz HM, Chen YT, Wang Y, Vaccarino V, Radford MJ, Horwitz RI. Predictors of readmission among elderly survivors of admission with heart failure. Am Heart J 2000;139:72–77. 29. Braunstein JB, Anderson GF, Gerstenblith G, Weller W, Niefeld M, Herbert R, Wu AW. Noncardiac comorbidity increases preventable hospitalizations and mortality among Medicare beneficiaries with chronic heart failure. J Am Coll Cardiol 2003;42:1226 –1233.
0002-9149/11/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2011.05.032
payer organizations are increasingly interested in improving care for patients with HF, decreasing readmissions, and lowering costs for Medicare beneficiaries. For example, the Centers for Medicare and Medicaid Services publicly reports hospital-level 30-day risk-standardized rates of readmission and mortality for HF.3,4 Early physician follow-up after discharge is a central component of efforts to decrease risk of readmission.5–7 Physician follow-up within 7 days of discharge decreases 30-day readmission rates in patients with HF.7 Despite mounting evidence supporting early followup, many patients do not receive it. Outpatient office visits are a supply-sensitive resource with often changing times to availability. Whether physician density or other associated factors affect access to early physician follow-up in recently discharged patients with HF is not well understood.8 Therefore, we examined data from the Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients With Heart Failure (OPTIMIZE-HF) registry and the Get With The Guidelines–Heart Failure (GWTG-HF) registry linked to inpatient and outpatient Medicare claims to describe patient-level characteristics and market factors associated with early physician follow-up after discharge from an HF hospitalization. Methods We linked Medicare inpatient claims data for January 1, 2003 through December 31, 2006 with data from the www.ajconline.org
986
The American Journal of Cardiology (www.ajconline.org)
Table 1 Baseline characteristics of study population Characteristic
Age (years) Age group (years) 65–69 70–74 75–79 ⱖ80 Women Race Black Other Medical history Anemia Atrial arrhythmia Chronic obstructive pulmonary disease Chronic renal insufficiency Coronary artery disease or ischemic heart disease Depression Diabetes mellitus Current or previous treatment for hyperlipidemia Current or previous treatment for hypertension Left ventricular systolic function Preserved systolic function Left ventricular systolic dysfunction Missing data Peripheral vascular disease Previous cerebrovascular accident or transient ischemic attack Smoker within the past year Findings on admission Hemoglobin (g/dl) Mean ⫾ SD Median (interquartile range) Serum creatinine (mg/dl) ⬍1.5 1.5–⬍2.0 ⱖ2.0 Serum sodium (mEq/L) Systolic blood pressure (mm Hg) Rural location State Medicaid buy-in Physicians per 100,000 residents Physicians per 100,000 residents in hospital referral region Quartile 1 (116–175) Quartile 2 (176–189) Quartile 3 (190–210) Quartile 4 (211–320) Year of index hospitalization 2003 2004 2005 2006
Early Follow-Up
p Value
Yes (n ⫽ 11,420)
No (n ⫽ 18,716)
79.0 (73.0–84.0)
79.0 (73.0–84.0)
0.03
1,460 (12.8%) 1,929 (16.9%) 2,596 (22.7%) 5,435 (47.6%) 5,858 (51.3%)
2,659 (14.2%) 3,364 (18.0%) 3,996 (21.4%) 8,697 (46.5%) 10,211 (54.6%)
⬍0.001 0.02 0.005 0.06 ⬍0.001
1,014 (8.9%) 10,406 (91.1%)
2,200 (11.8%) 16,516 (88.2%)
⬍0.001 ⬍0.001
1,928 (16.9%) 4,342 (38.0%) 3,014 (26.4%) 1,991 (17.4%) 6,171 (54.0%) 989 (8.7%) 4,464 (39.1%) 4,359 (38.2%) 8,148 (71.3%)
3,096 (16.5%) 6,350 (33.9%) 5,256 (28.1%) 3,464 (18.5%) 9,937 (53.1%) 1,691 (9.0%) 7,428 (39.7%) 6,754 (36.1%) 13,537 (72.3%)
0.44 ⬍0.001 0.001 0.02 0.11 0.27 0.30 ⬍0.001 0.07
5,837 (51.1%) 4,201 (36.8%) 1,382 (12.1%) 1,519 (13.3%) 1,794 (15.7%) 1,007 (8.8%)
9,146 (48.9%) 7,291 (39.0%) 2,279 (12.2%) 2,656 (14.2%) 3,003 (16.0%) 1,908 (10.2%)
⬍0.001 ⬍0.001 0.85 0.03 0.44 ⬍0.001 0.15
12.2 ⫾ 8.7 12.0 (10.7–13.4) 1.3 (1.0–1.8) 6,786 (59.4%) 2,386 (20.9%) 2,173 (19.0%) 138.0 (135.0–141.0) 140.0 (121.0–160.0) 1,940 (17.0%) 1,683 (14.7%) 191.5 (183.1–210.4)
12.1 ⫾ 3.7 12.1 (10.8–13.4) 1.3 (1.0–1.8) 10,833 (57.9%) 3,820 (20.4%) 3,930 (21.0%) 138.0 (136.0–141.0) 140.0 (120.0–160.0) 3,688 (19.7%) 3,510 (18.8%) 191.4 (177.8–210.0)
1,557 (13.6%) 3,445 (30.2%) 3,583 (31.4%) 2,835 (24.8%)
3,031 (16.2%) 5,761 (30.8%) 5,542 (29.6%) 4,382 (23.4%)
⬍0.001 0.008 0.32 ⬍0.001 ⬍0.001 0.98 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 0.26 0.001 0.005
2,676 (23.4%) 4,269 (37.4%) 1,741 (15.2%) 2,734 (23.9%)
4,608 (24.6%) 6,892 (36.8%) 2,909 (15.5%) 4,307 (23.0%)
0.02 0.33 0.49 0.07
Values are presented as median (interquartile range), number of patients (percentage), or mean ⫾ SD.
OPTIMIZE-HF and GWTG-HF registries. GWTG-HF is a continuation of the OPTIMIZE-HF under the sponsorship of the American Heart Association, and the registries had the same design, inclusion criteria, and data collection methods.9,10 Patients were eligible for inclusion in the registries if they were admitted to a hospital for an episode of worsening HF or developed significant HF symptoms during a
hospitalization for which HF was the primary discharge diagnosis. Participating institutions submitted data on consecutive eligible patients in compliance with Joint Commission and Centers for Medicare and Medicaid Services standards. The registries included hospitals from all regions of the United States ranging in type from community hospitals to academic tertiary care referral centers. The validity and
Heart Failure/Early Physician Follow-up for Heart Failure
987
Table 2 Inpatient procedures, discharge processes, and discharge medications Variable
Percutaneous coronary intervention performed during hospitalization Coronary artery bypass grafting performed during hospitalization Automated implantable cardioverter–defibrillator placement in hospitalization Right cardiac catheterization performed during hospitalization Length of stay (days), median (interquartile range) Referral to outpatient heart failure management program Discharge instructions completed  Blocker Angiotensin-converting enzyme inhibitor or angiotensin receptor blocker Antiplatelet agent Lipid-lowering agent Digoxin Diuretic Aldosterone antagonist
generalizability of the OPTIMIZE-HF registry have been described previously.11 We obtained research-identifiable Medicare inpatient, carrier, and denominator files for each patient in the study for 2003 through 2007. Denominator files contain death date and Medicare eligibility dates. We used the inpatient files to identify readmissions and the carrier files to examine early physician follow-up after discharge from the index HF hospitalization. Carrier files contain claims from noninstitutional medical providers for services covered by Medicare Part B and include Healthcare Common Procedure Coding System codes, physician specialty codes, and service dates. We used data from 2007 for hospital discharges that occurred in December 2006. This study was approved by the institutional review board of the Duke University Health System. All sites participating in OPTIMIZE-HF and GWTG-HF received local institutional review board approval, if applicable, and complied with all local regulatory and privacy guidelines. Outcome Sciences, Inc. (Cambridge, Massachusetts) served as the data collection and coordination center for GWTGHF. The Duke Clinical Research Institute (Durham, North Carolina) served as the data analysis center and entered an agreement to analyze the aggregate deidentified data for research purposes (http://www.Clinicaltrials.gov, identifier NCT00344513). We used a method that links registry records for patients ⱖ65 years with inpatient Medicare claims files based on indirect identifiers.10 Using this method we were able to link 62,311 (78%) of the 79,837 eligible hospitalizations in the OPTIMIZE-HF and GWTG-HF registries with Medicare inpatient claims. Eligible patients were enrolled in fee-forservice Medicare and were discharged alive from a fully participating site in the OPTIMIZE-HF or GWTG-HF program. For patients with multiple admissions we selected the first as the index admission. We excluded patients who were discharged to a skilled nursing facility (n ⫽ 9,166) or to hospice care (n ⫽ 804). We also excluded 1,390 patients from 143 hospitals that had ⬍25 eligible patients remaining in the cohort.
Early Follow-Up Yes (n ⫽ 11,420)
No (n ⫽ 18,716)
161 (1.4%) 56 (0.5%) 311 (2.7%) 307 (2.7%) 4 (2–6) 1,453 (12.7%) 6,959 (60.9%) 7,859 (68.8%) 7,316 (64.1%) 6,063 (53.1%) 5,005 (43.8%) 2,988 (26.2%) 9,309 (81.5%) 1,534 (13.4%)
243 (1.3%) 123 (0.7%) 623 (3.3%) 534 (2.9%) 4 (3–6) 2,166 (11.6%) 11,179 (59.7%) 12,612 (67.4%) 11,750 (62.8%) 10,468 (55.9%) 7,861 (42.0%) 4,807 (25.7%) 14,784 (79.0%) 2,336 (12.5%)
p Value
0.41 0.07 0.003 0.40 ⬍0.001 0.003 0.04 0.01 0.03 ⬍0.001 0.002 0.36 ⬍0.001 0.02
We defined early physician follow-up as an outpatient evaluation and management visit with a physician (Healthcare Common Procedure Coding System codes 992.xx through 994.xx) within 7 days after discharge from the index hospitalization. Although early follow-up can be defined from any number of time points, we chose 7 days to be consistent with the American College of Cardiology’s H2H national quality-improvement initiative. The H2H initiative involves ⬎800 centers in an effort to decrease hospital readmissions in patients with HF and myocardial infarction by 20% by 2012. Postdischarge physician follow-up within 7 days is 1 of 3 core goals of the initiative.6 The importance of physician follow-up within 7 days is also supported by evidence from a recent study in which hospitals that arranged 7-day follow-up for a larger proportion of patients with HF had lower readmission rates than other hospitals.7 We excluded emergency department visits from calculations of early follow-up rates because they were unplanned visits not reflecting a system of care. We obtained patient demographic characteristics, medical history, results of admission laboratory tests and examinations, discharge pharmacy records, and procedural information for the index hospitalization from the registry data. Patients were assigned to race categories using options available on the case-report form. We used the reported category “black” and combined all others as “other.” Variables in the analysis had low rates of “missingness” (i.e., ⬍5% of records) with the exception of left ventricular function (12.1%). For continuous variables and the variable for left ventricular function, we created categorical variables that included a category for missing values. For other dichotomous variables (i.e., smoker within previous year, discharge processes and performance measurements, and index hospitalization procedures) we imputed missing values to “no.” We used the enrollment code in the month of the index hospitalization to ascertain Medicaid eligibility, a marker of socioeconomic status.12,13 A dichotomous indicator variable for rural status was derived from rural-urban commuting area scores based on zip code of residence14 using the University of Washington classification C algo-
988
The American Journal of Cardiology (www.ajconline.org)
Table 3 Univariate and multivariable predictors of seven-day physician follow-up Variable Age group (years) 65–69 70–74 75–79 ⱖ80 Female gender Black race Medical history Atrial arrhythmia Chronic obstructive pulmonary disease Coronary artery disease or ischemic heart disease Depression Diabetes mellitus Hyperlipidemia Peripheral vascular disease Previous cerebrovascular accident or transient ischemic attack Smoker within previous year Systolic function Preserved systolic function Left ventricular systolic dysfunction Findings on admission Serum creatinine (mg/dl) ⬍1.5 1.5–⬍2.0 ⱖ2.0 Systolic blood pressure (mm Hg) ⬍120 120–⬍140 140–⬍160 ⱖ160 Missing Serum sodium (mEq/L) ⬍135 135–⬍145 ⱖ145 Hemoglobin (g/dl) ⬍9 9–⬍12 ⱖ12 Findings at discharge Referral to outpatient heart failure management program Discharge instructions completed Length of index hospitalization ⬎7 days Rural location State Medicaid buy-in Quartile of physicians per 100,000 residents in hospital referral region 1 (116–175) 2 (176–189) 3 (190–210) 4 (211–320) Year of index hospitalization 2003 2004 2005 2006 * Multivariable model includes all listed variables. CI ⫽ confidence interval; OR ⫽ odds ratio.
Unadjusted OR (95% CI)
p Value
Adjusted OR* (95% CI)
p Value
1.00 (reference) 1.04 (0.96–1.13) 1.18 (1.08–1.28) 1.11 (1.03–1.20) 0.88 (0.84–0.92) 0.76 (0.70–0.83)
0.35 ⬍0.001 0.005 ⬍0.001 ⬍0.001
1.00 (reference) 1.01 (0.93–1.11) 1.12 (1.03–1.22) 1.04 (0.96–1.13) 0.87 (0.83–0.91) 0.84 (0.77–0.92)
0.74 0.007 0.29 ⬍0.001 ⬍0.001
1.20 (1.14–1.26) 0.94 (0.89–0.99) 1.05 (1.00–1.10) 0.98 (0.90–1.07) 0.99 (0.94–1.04) 1.10 (1.05–1.16) 0.94 (0.88–1.01) 0.98 (0.92–1.05)
⬍0.001 0.02 0.07 0.68 0.62 ⬍0.001 0.12 0.53
1.17 (1.11–1.23) 0.94 (0.89–1.00) 1.03 (0.98–1.09) 0.99 (0.91–1.08) 1.01 (0.96–1.06) 1.09 (1.04–1.15) 0.94 (0.87–1.00) 0.99 (0.92–1.05)
⬍0.001 0.04 0.21 0.87 0.68 0.001 0.06 0.68
0.87 (0.80–0.94)
0.001
0.92 (0.84–1.01)
0.07
1.00 (reference) 0.91 (0.87–0.96)
0.001
1.00 (reference) 0.89 (0.85–0.94)
⬍0.001
1.00 (reference) 1.00 (0.94–1.06) 0.89 (0.84–0.95)
0.92 ⬍0.001
1.00 (reference) 0.97 (0.91–1.03) 0.88 (0.82–0.94)
0.35 ⬍0.001
1.00 (reference) 1.02 (0.95–1.09) 1.01 (0.95–1.09) 1.02 (0.95–1.09) 1.11 (0.75–1.65)
0.62 0.72 0.59 0.59
1.00 (reference) 1.02 (0.95–1.09) 1.02 (0.95–1.10) 1.05 (0.98–1.13) 1.00 (0.67–1.52)
0.66 0.54 0.19 0.98
1.00 (reference) 0.90 (0.85–0.96) 0.88 (0.76–1.01)
0.001 0.06
1.00 (reference) 0.90 (0.85–0.96) 0.89 (0.77–1.02)
0.001 0.09
1.00 (reference) 1.07 (0.96–1.20) 1.02 (0.91–1.13)
0.23 0.77
1.00 (reference) 1.05 (0.94–1.17) 0.97 (0.87–1.08)
0.40 0.59
1.16 (1.07–1.25)
⬍0.001
1.14 (1.05–1.24)
0.002
1.07 (1.01–1.13) 0.96 (0.90–1.02) 0.84 (0.78–0.90) 0.74 (0.69–0.79)
0.01 0.19 ⬍0.001 ⬍0.001
1.04 (0.98–1.10) 0.97 (0.91–1.04) 0.84 (0.78–0.91) 0.79 (0.74–0.85)
0.20 0.38 ⬍0.001 ⬍0.001
1.00 (reference) 1.23 (1.08–1.41) 1.29 (1.14–1.47) 1.33 (1.16–1.52)
0.002 ⬍0.001 ⬍0.001
1.00 (reference) 1.21 (1.06–1.38) 1.29 (1.14–1.46) 1.29 (1.12–1.48)
0.005 ⬍0.001 ⬍0.001
1.00 (reference) 1.04 (0.97–1.12) 1.00 (0.91–1.10) 1.12 (1.02–1.23)
0.22 0.99 0.02
1.00 (reference) 1.03 (0.96–1.10) 0.99 (0.89–1.09) 1.09 (0.99–1.20)
0.46 0.77 0.08
Heart Failure/Early Physician Follow-up for Heart Failure
rithm.15 We assigned each patient to a hospital referral region according to the zip code of residence.16 We assigned hospital referral regions to quartiles based on total number of physicians per 100,000 residents and linked rankings to patients by hospital referral region. For baseline patient characteristics and index hospitalization measurements we present categorical variables as frequencies with percentages and continuous variables as medians with interquartile ranges. To test for differences in early physician follow-up, we used chi-square tests for categorical variables and Kruskal–Wallis tests for continuous variables. We used univariate logistic regression models and multivariable logistic regression models with site-level random effects to examine predictors of early physician follow-up. In multivariable analysis we modeled early follow-up as a function of age, gender, race, medical history, results of admission laboratory tests and examinations, completion of discharge instructions, referral to an HF disease management program, length of stay ⬎7 days for the index hospitalization, rural location, state Medicaid buy-in, quartile of physicians per 100,000 residents in the hospital referral region, and year of index hospitalization. We used a similar approach to examine unadjusted relations between covariates and early physician follow-up. In the primary analysis we included all patients in the study population. In a sensitivity analysis we excluded patients who died or were readmitted within 7 days after discharge from the index hospitalization. In a second sensitivity analysis we determined unadjusted and adjusted relations between covariates and early physician follow-up defined as 14 days including and excluding patients who died or were readmitted at 14 days. We used SAS 9.2 (SAS Institute, Cary, North Carolina) for all analyses. Results The study population included 30,136 patients from 225 hospitals. Within 7 days of discharge from the index HF hospitalization, 11,420 patients (37.9%) had a follow-up visit with a physician. Table 1 lists baseline demographic characteristics of the study population stratified by early follow-up status. Median age in the 2 groups was 79 years. Compared to the early follow-up cohort, the cohort of patients without early follow-up had a larger proportion of black patients and women. The early follow-up cohort had higher rates of preserved systolic function and history of atrial arrhythmias. The early follow-up cohort also had lower rates of chronic obstructive pulmonary disease and chronic renal insufficiency. A larger proportion of patients with HF in the early follow-up cohort resided in a hospital referral region with a higher concentration of physicians. More patients in the cohort without early follow-up lived in a rural location and had lower socioeconomic status. Table 2 presents inpatient procedures, discharge process measurements, and discharge medications stratified by early follow-up status. In general the 2 cohorts were similar with respect to findings at discharge and procedures performed. However, larger proportions of patients in the early follow-up group were referred to an outpatient HF disease management program and had discharge instructions completed. The early follow-up cohort also had higher rates of
989
discharge with evidence-based HF medications including angiotensin-converting enzyme inhibitors or angiotensin receptor blockers,  blockers, and aldosterone antagonists. A larger proportion of the early follow-up cohort was discharged on diuretics and lipid-lowering agents. The early follow-up group had a smaller proportion of patients discharged with antiplatelet medications. Table 3 lists unadjusted and adjusted associations with early physician follow-up. After adjustment for baseline characteristics of the index hospitalization, we observed increased odds of early follow-up by quartile of number of physicians per 100,000 patients in the hospital referral region. Patients residing in hospital referral regions in the top quartile of physician density had greater odds of early follow-up compared to patients residing in regions in the lowest quartile. Moreover, patients living in rural areas and patients with lower socioeconomic status were less likely to have early physician follow-up. We found no significant improvement in crude rates or adjusted odds of early follow-up from 2003 through 2006. After multivariable adjustment for baseline characteristics of the study population, the odds of early follow-up were 13% lower in women compared to men and 16% lower in black patients compared to patients of other races. In addition, we found statistically significant independent associations between early follow-up and history of hyperlipidemia, history of atrial arrhythmias, and referral to an outpatient HF disease management program. Patients with left ventricular dysfunction, chronic obstructive pulmonary disease, or chronic renal insufficiency had lower odds of early follow-up. In a sensitivity analysis we repeated the primary analysis but excluded patients from the cohort who died or were readmitted within 7 days of discharge from the index hospitalization. This analysis did not significantly change the findings for predictors of early follow-up. In a second sensitivity analysis we changed our definition of early physician follow-up to 14 days after discharge including and excluding the population of patients who died or were readmitted at 14 days. Overall the associations and effect sizes were similar to the 7-day model. There were 2 exceptions. First, length of stay ⬎7 days was associated with decreased odds of early follow-up in the model that included patients who died or were readmitted within 14 days. Second, completed discharge instructions were associated with increased odds of early follow-up. Discussion To our knowledge, ours is the largest analysis to date of factors associated with early physician follow-up after discharge from an HF admission. The study yielded several important findings. First, patients who reside in hospital referral regions with low physician density were less likely to receive early physician follow-up. Second, overall rate of early physician follow-up was low (38%). Third, gender and race were associated with early follow-up. Fourth, patients at high risk for readmission such as those with kidney disease and chronic obstructive pulmonary disease were less likely to receive early follow-up.
990
The American Journal of Cardiology (www.ajconline.org)
The definition of early physician follow-up as 7 days after discharge is consistent with current quality-improvement initiatives and previous research. However, to determine whether the same factors were associated with early physician follow-up if we used a different cut point, we repeated the analysis using a 14-day cut point instead of 7 days and found similar associations and effect sizes. Our finding of higher rates of early follow-up in patients residing in hospital referral regions with more physicians per capita calls attention to provider access as a factor that influences the transition from inpatient to outpatient care. To our knowledge this is the first study to identify a correlation between physician supply and access to HF care. One might argue that physician density is simply a marker for something else—notably, rural location or socioeconomic status. However, we adjusted for rural location and socioeconomic status (based on Medicaid buy-in) and physician density remained significant. Moreover, patients in rural regions and from a lower socioeconomic status were less likely to receive early physician follow-up. Other studies have found geographic variations in subspecialty physician supply.17 For example, areas with lower physician density have lower rates of primary care visits and early follow-up for newborns.18 Whether physician density is directly associated with lower rates of follow-up or is a marker of other socioeconomic factors is uncertain. A recent study used a multivariable model to determine predictors of myocardial infarction and HF hospitalization rates. After adjustment for other socioeconomic variables (i.e., unemployment rate, poverty rate, median income, rural location, distance to hospital, and education level), physician density was not a significant predictor.19 However, outpatient physician visits are supply-sensitive and are thought to play a role in variations in regional health care resource use.8 Providers with larger patient panels will have a more difficult time fitting patients into their clinic after a hospital discharge in a timely fashion. In comparison, physicians with more clinic slots to fill have an incentive to schedule more patients.20 Whether increased supply-sensitive care generally for patients with HF improves overall outcome is unclear. However, our analysis suggests that increased provider access may promote early physician follow-up, which has been associated with decreased short-term readmission rates. Consistent with previous studies in other areas of medicine, black race was associated with lower odds of early physician follow-up.21–24 This finding may help to explain why previous research has shown a greater risk of readmission in black patients after hospitalization for HF.2,25,26 Similarly, women in our study were less likely to have early physician follow-up, which may contribute to higher risks for HF readmission in women.24 Therefore, improving transitional care and access to follow-up for these groups requires increased attention. We also found that patients who were at greater clinical risk for readmission or death because of left ventricular systolic dysfunction, increased serum creatinine, or chronic obstructive pulmonary disease were less likely to receive early follow-up.27–29 The explanation for this seemingly paradoxical finding of lower rates of early follow-up among subsets of higher-risk more complicated patients is unclear.
Sicker patients may have more difficulty arranging physician visits or may tend to be referred to specialists with longer wait times for visits. It is also possible that higherrisk patients with more co-morbid conditions are more likely to be readmitted or to die within 7 days of discharge and therefore do not have an opportunity to present for early follow-up. However, findings from the sensitivity analysis that excluded patients who died or were readmitted within 7 days of discharge were similar to findings from the primary analysis. Atrial arrhythmias stood out as a co-morbid condition that resulted in greater odds of early follow-up, possibly related to a need in these patients for early frequent international normalized ratio testing after discharge and physician awareness of the potentially catastrophic consequences of subtherapeutic or supratherapeutic international normalized ratio levels. We also did not find a meaningful temporal change in rates of early follow-up and year of the index hospitalization was not an independent predictor of early follow-up. This finding highlights 1 of several domains in the care of patients with acute HF who have not shown improvement over time. Our finding that enrollment in an HF disease management program after discharge is associated with early physician follow-up offers a potential intervention for the aforementioned risk/follow-up paradox. Enrollment in a disease management program may provide closer monitoring of sicker patients, thus improving adherence to follow-up appointments. Further study of who benefits most from such programs will be important. Our study has several limitations. First, analysis was restricted to patients in hospitals participating in qualityimprovement registries, which may not be generalizable to other hospitals. A previous analysis comparing registry data to the larger population found the populations to be similar.11 Second, the study included only fee-for-service Medicare beneficiaries and may not generalize to patients enrolled in Medicare managed care. Third, patients may have received follow-up care from nurse practitioners, physician assistants, home health visits, or remote monitoring that we did not capture in this analysis. Fourth, the analysis did not account for unmeasured confounders that may influence access to care. Acknowledgment: Damon M. Seils, MA (Duke University) provided editorial assistance and prepared the report. Mr. Seils did not receive compensation for his assistance apart from his employment at the institution where the study was conducted. 1. Lloyd-Jones D, Adams RJ, Brown TM, Carnethon M, Dai S, De Simone G, Ferguson TB, Ford E, Furie K, Gillespie C, Go A, Greenlund K, Haase N, Hailpern S, Ho PM, Howard V, Kissela B, Kittner S, Lackland D, Lisabeth L, Marelli A, McDermott MM, Meigs J, Mozaffarian D, Mussolino M, Nichol G, Roger VL, Rosamond W, Sacco R, Sorlie P, Roger VL, Thom T, Wasserthiel-Smoller S, Wong ND, Wylie-Rosett J; American Heart Association, Statistics Committee and Stroke Statistics. Heart disease and stroke statistics—2010 update: a report from the American Heart Association. Circulation 2010;121(suppl):e46 – e215. 2. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009; 360:1418 –1428.
Heart Failure/Early Physician Follow-up for Heart Failure 3. Medicare Payment Advisory Commission (MedPAC). Report to the Congress: promoting greater efficiency in Medicare, June 2007. Available at: http://www.medpac.gov/documents/Jun07_EntireReport.pdf. Accessed August 25, 2010. 4. US Department of Health and Human Services. Hospital compare. Available at: http://www.hospitalcompare.hhs.gov. Accessed December 6, 2009. 5. Hauptman PJ, Rich MW, Heidenreich PA, Chin J, Cummings N, Dunlap ME, Edwards ML, Gregory D, O’Connor CM, Pezzella SM, Philbin E; Heart Failure Society of America. The heart failure clinic: a consensus statement of the Heart Failure Society of America. J Card Fail 2008;14:801– 815. 6. H2H National Quality Improvement Initiative. Available at: http:// h2hquality.org. Accessed December 26, 2009. 7. Hernandez AF, Greiner MA, Fonarow GC, Hammill BG, Heidenreich PA, Yancy CW, Peterson ED, Curtis LH. Relationship between early physician follow-up and 30-day readmission among Medicare beneficiaries hospitalized for heart failure. JAMA 2010;303:1716 –1722. 8. Fisher ES, Wennberg JE. Health care quality, geographic variations, and the challenge of supply-sensitive care. Perspect Biol Med 2003; 46:69 –79. 9. Fonarow GC, Abraham WT, Albert NM, Gattis WA, Gheorghiade M, Greenberg B, O’Connor CM, Yancy CW, Young J. Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF): rationale and design. Am Heart J 2004;148:43–51. 10. Hammill BG, Hernandez AF, Peterson ED, Fonarow GC, Schulman KA, Curtis LH. Linking inpatient clinical registry data to Medicare claims data using indirect identifiers. Am Heart J 2009;157:995–1000. 11. Curtis LH, Greiner MA, Hammill BG, DiMartino LD, Shea AM, Hernandez AF, Fonarow GC. Representativeness of a national heart failure quality-of-care registry: comparison of OPTIMIZE-HF and non-OPTIMIZE-HF Medicare patients. Circ Cardiovasc Qual Outcomes 2009;2:377–384. 12. Section 3. Dual-eligible beneficiaries. In: Medicare Payment Advisory Commission (MedPAC). A Data Book: Healthcare Spending and the Medicare Program, June 2010. Available at: http://www.medpac.gov/ chapters/Jun10DataBookSec3.pdf. Accessed March 31, 2011. 13. Section 8. How do dual eligible Medicare Benes compare to non-dual eligible Medicare Benes? In: Medicare Current Beneficiary Survey (MCBS) Data Tables, 1997. Available at: http://www.cms.gov/MCBS/ Downloads/CNP_1997_summary8.pdf. Accessed March 31, 2011. 14. US Department of Agriculture, Economic Research Service. Ruralurban commuting area codes. Available at: http://www.ers.usda.gov/ Data/RuralUrbanCommutingAreaCodes/. Accessed March 31, 2011. 15. WWAMI Rural Health Research Center. RUCA data: using RUCA data. Available at: http://depts.washington.edu/uwruca/ruca-uses.php. Accessed March 31, 2011. 16. The Dartmouth Atlas of Health Care. Available at: http://www. dartmouthatlas.org/data/download.shtm. Accessed February 18, 2010.
991
17. Onega T, Duell EJ, Shi X, Wang D, Demidenko E, Goodman D. Geographic access to cancer care in the U.S. Cancer 2008;112:909 – 918. 18. Guttmann A, Shipman SA, Lam K, Goodman DC, Stukel TA. Primary care physician supply and children’s health care use, access, and outcomes: findings from Canada. Pediatrics 2010;125:1119 –1126. 19. Harris DE, Aboueissa AM, Hartley D. Myocardial infarction and heart failure hospitalization rates in Maine, USA—variability along the urban-rural continuum. Rural Remote Health 2008;8:980. 20. Wennberg JE. Unwarranted variations in healthcare delivery: implications for academic medical centres. BMJ 2002;325:961–964. 21. Sonel AF, Good CB, Mulgund J, Roe MT, Gibler WB, Smith SC, Jr., Cohen MG, Pollack CV Jr, Ohman EM, Peterson ED, CRUSADE Investigators; CRUSADE Investigators. Racial variations in treatment and outcomes of black and white patients with high-risk non-STelevation acute coronary syndromes: insights from CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines?). Circulation 2005;111:1225–1232. 22. Skinner J, Chandra A, Staiger D, Lee J, McClellan M. Mortality after acute myocardial infarction in hospitals that disproportionately treat black patients. Circulation 2005;112:2634 –2641. 23. Blomkalns AL, Chen AY, Hochman JS, Peterson ED, Trynosky K, Diercks DB, Brogan GX, Jr., Boden WE, Roe MT, Ohman EM, Gibler WB, Newby LK; CRUSADE Investigators. Gender disparities in the diagnosis and treatment of non–ST-segment elevation acute coronary syndromes: large-scale observations from the CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes with Early Implementation of the American College of Cardiology/American Heart Association Guidelines) National Quality Improvement Initiative. J Am Coll Cardiol 2005;45:832– 837. 24. Howie-Esquivel J, Dracup K. Effect of gender, ethnicity, pulmonary disease, and symptom stability on rehospitalization in patients with heart failure. Am J Cardiol 2007;100:1139 –1144. 25. Philbin EF, DiSalvo TG. Influence of race and gender on care process, resource use, and hospital-based outcomes in congestive heart failure. Am J Cardiol 1998;82:76 – 81. 26. Rathore SS, Foody JM, Wang Y, Smith GL, Herrin J, Masoudi FA, Wolfe P, Havranek EP, Ordin DL, Krumholz HM. Race, quality of care, and outcomes of elderly patients hospitalized with heart failure. JAMA 2003;289:2517–2524. 27. Patel UD, Greiner MA, Fonarow GC, Phatak H, Hernandez AF, Curtis LH. Associations between worsening renal function and 30-day outcomes among Medicare beneficiaries hospitalized with heart failure. Am Heart J 2010;160(suppl):132.e1–138.e1. 28. Krumholz HM, Chen YT, Wang Y, Vaccarino V, Radford MJ, Horwitz RI. Predictors of readmission among elderly survivors of admission with heart failure. Am Heart J 2000;139:72–77. 29. Braunstein JB, Anderson GF, Gerstenblith G, Weller W, Niefeld M, Herbert R, Wu AW. Noncardiac comorbidity increases preventable hospitalizations and mortality among Medicare beneficiaries with chronic heart failure. J Am Coll Cardiol 2003;42:1226 –1233.